University of Birmingham

Construction grammar and the corpus-basedanalysis of discoursesGroom, Nicholas

DOI:10.1075/ijcl.00014.gro

License:None: All rights reserved

Document VersionPeer reviewed version

Citation for published version (Harvard):Groom, N 2019, 'Construction grammar and the corpus-based analysis of discourses: the case of the WAY INWHICH construction', International Journal of Corpus Linguistics, vol. 24, no. 3, pp. 291-323.https://doi.org/10.1075/ijcl.00014.gro

Link to publication on Research at Birmingham portal

Publisher Rights Statement:Checked for eligibility: 22/05/2019

Groom, N. (2019) Construction grammar and the corpus-based analysis of discourses: the case of the WAY IN WHICH construction,International Journal of Corpus Linguistics, 24(3), 335–367, https://doi.org/10.1075/ijcl.00014.gro

© John Benjamins Publishing Company

General rightsUnless a licence is specified above, all rights (including copyright and moral rights) in this document are retained by the authors and/or thecopyright holders. The express permission of the copyright holder must be obtained for any use of this material other than for purposespermitted by law.

•Users may freely distribute the URL that is used to identify this publication.•Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of privatestudy or non-commercial research.•User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?)•Users may not further distribute the material nor use it for the purposes of commercial gain.

Where a licence is displayed above, please note the terms and conditions of the licence govern your use of this document.

When citing, please reference the published version.

Take down policyWhile the University of Birmingham exercises care and attention in making items available there are rare occasions when an item has beenuploaded in error or has been deemed to be commercially or otherwise sensitive.

If you believe that this is the case for this document, please contact UBIRA@lists.bham.ac.uk providing details and we will remove access tothe work immediately and investigate.

Download date: 27. Jan. 2022

1

Construction Grammar and the corpus-based analysis of discourses:

The case of the WAY IN WHICH construction

Nicholas Groom

Department of English Language and Linguistics,

University of Birmingham

Construction grammar (CxG) initially arose as a usage-based alternative to nativist theoretical

accounts of language, and remains to this day strongly associated with cognitive linguistic

theory and research. In this paper, however, I argue that CxG can be seen as offering an

equally viable general framework for socially-oriented linguists whose work focuses on the

corpus-based analysis of discourses (CBADs). The paper begins with brief reviews of CxG

and CBADs as distinctive research traditions, before going on to identify synergies (both

potential and actual) between them. I then offer a more detailed case study example, focusing

on a usage-based analysis of a newly identified construction, the WAY IN WHICH

construction, as it occurs in corpora representing six different academic discourses. The paper

concludes by rebutting some anticipated objections to the approach advocated here, and by

proposing a new conceptual model for constructionist approaches to CBADs.

Keywords: Construction Grammar, discourse analysis, discourses, phraseology

1. Introduction

Since it first emerged in the 1980s, Construction Grammar has consistently been associated

with cognitive linguistics as a general theoretical enterprise (cf. Lakoff, 1987; Fillmore et al.,

1988; Croft & Cruse, 2004; Evans & Green, 2006; Hoffmann & Trousdale, 2013) and with

cognitively-oriented research on topics such as first and second language acquisition (e.g.

Goldberg, 1995, 2006; Ellis, 2003; Gries & Wulff, 2005; Gilquin, 2015; Römer et al., 2015;

Kerz & Wiechmann, 2016; Ellis et al., 2016) and language change (e.g. Patten, 2012; Hilpert,

2013; Traugott & Trousdale, 2013; Jing-Schmidt & Peng, 2016; Perek, 2016). At the same

time, however, there has also been a steady growth of interest in extending Construction

Grammar into the more socially and contextually oriented fields of sociolinguistics,

pragmatics and discourse analysis (e.g. Fried & Östman, 2005; Bergs & Diewald, 2009;

2

Östman & Trousdale, 2013; Aijmer, 2016; Nikiforidou & Fischer, 2016). The aim of this

paper is to promote the further extension of Construction Grammar into these realms, by

making a practical and theoretical case for its applicability to the now burgeoning subfield of

corpus-based research that focuses on the analysis of discourses, defined as constellations of

meanings and values which simultaneously serve to represent things in the world in particular

ways, and to represent the social actors who express these meanings and values as members

of particular sociocultural groups (cf. Teubert, 2005, 2010; Baker, 2006; Partington et al.,

2013; Baker & McEnery, 2015; Taylor & Marchi, 2018).

In order to establish the terms and the remit of the argument in more detail, the paper

begins with brief theoretical overviews of the two research traditions in question, and of

previous interfaces between them. I then offer a detailed case study example analysis

focusing on a newly identified construction, which I shall refer to as the WAY IN WHICH

construction (e.g. this affected the way in which their food was cooked and served; from the

BNC), in corpora representing six different academic discourses. The paper concludes with a

critical discussion of some anticipated objections to the approach advocated here.

2. Construction Grammar

As Construction Grammar (conventionally abbreviated to CxG) is not (yet) a mainstream

approach within the field of discourse analysis, it will be necessary to begin with a very brief

outline of this approach for the benefit of readers who may be unfamiliar with its main tenets.

The first point to note about CxG is that, just like the Chomskyan nativist models of

language to which it is in almost all other respects diametrically opposed, it is not simply a

descriptive grammar but a fully-fledged attempt to address the fundamental theoretical

question ‘what is language?’ CxG’s answer to this question is that each human language is an

intricately structured, semantically motivated and constantly evolving inventory of symbolic

units, referred to as constructions. Each construction is symbolic in that it comprises a pairing

of a particular form with a particular meaning. These form-meaning pairs are of greatly

varying size and schematicity, ranging from single morphemes such as dog to clause-length

units such as you can’t teach an old dog new tricks, and from fully lexicalized expressions

such as you can’t teach an old dog new tricks to entirely schematic sequences such as

SUBJagent V OBJrecipient OBJtheme (i.e. the English ditransitive construction, of which the

proverb you can’t teach an old dog new tricks constitutes a very particular exemplar). While

3

it may be convenient for certain purposes to talk informally about ‘words’ or ‘grammatical

structures’ as though they are fundamentally different things, such category distinctions have

no ontological status in CxG. On the contrary, ‘words’, ‘grammatical structures’ and

everything in between are understood as essentially the same kind of unit, differing only in

terms of length, complexity or level of generality. CxG is thus a rigorously monist theory of

language, which insists that “knowledge of language is to be modelled as knowledge of

constructions, and nothing else in addition” (Hilpert, 2014: 22, emphasis added).

Furthermore, in answering the related question of how knowledge of a first or subsequent

language is acquired, CxG does not postulate any genetically endowed ‘language faculty’ of

the kind envisaged by Chomskyan generativism. On the contrary, CxG proposes that

language acquisition is motivated by social interaction, and facilitated by general learning

mechanisms such as the ability to perceive meaningful patterns, regularities and

correspondences in perceptual data and to infer productive generalisations from such

observations.

While these basic principles are shared by all construction grammarians, it is also

important to note that ‘Construction Grammar’ is not a single unified theoretical approach,

but rather a family of approaches and perspectives, each of which has its own distinctive

emphases and priorities (cf. Croft & Cruse, 2004; Hoffman & Trousdale, 2013). From the

perspective of the current paper, the most important contrast is that between ‘usage-based’

approaches such as Cognitive Construction Grammar (CCxG) (Goldberg, 1996) on the one

hand, and more ‘formal’ approaches such as Berkeley Construction Grammar (BCG)

(Fillmore & Kay, 1995; Kay, 2002; Fillmore, 2013) and Sign-Based Construction Grammar

(SBCG) (Boas & Sag, 2012; Michaelis, 2013) on the other. A helpful summary of the most

important differences between these two broad traditions is provided by Boas (2013: 248-49):

Whereas in CCxG the existence of constructions in the grammar [is] thought to be motivated

by more general properties of human interaction and cognition, BCG and SBCG do not

emphasize the role of motivation. [Instead], most work in BCG and SBCG is aimed at finding

maximal generalizations without any redundancy. This means that if a particular expression

can be accounted for on the basis of constructions already known to exist, then there is no

need to postulate a separate construction. Similarly, BCG and SBCG are not concerned with

frequencies of constructions. In contrast, CCxG takes a strong usage-based view of the role of

frequency and the status of item-specific instances, leading to the idea that even fully regular

4

patterns may be stored alongside more abstract schematic constructions when they occur with

sufficient frequency[.]

Given their commitment to accounting for observations about natural language usage

emerging from cognate fields such as psycholinguistics and corpus linguistics, it is perhaps

unsurprising to note that usage-based approaches (and CCxG in particular) have become the

default choice among corpus-based construction grammarians in recent years (Yoon & Gries,

2016). Accordingly, the usage-based definition of constructions proposed by Goldberg (2006:

5) will be assumed for the purposes of this paper:

Any linguistic pattern is recognized as a construction as long as some aspect of its form or

function is not strictly predictable from its component parts or from other constructions

recognized to exist. In addition, patterns are stored as constructions even if they are fully

predictable as long as they occur with sufficient frequency[.]

The final proviso specified in this definition – that constructions can be identified as such

purely on the basis of their frequency of occurrence – is of particular utility and relevance to

discourse researchers, as will be argued in more detail later in this paper.

3. The corpus-based analysis of discourses

Although the terms ‘discourse’ and ‘discourse analysis’ are notorious for meaning different

things to different scholars, it is possible to make a broad distinction between three different

(although ultimately interrelated) traditions of discourse research. One of these traditions (let

us label it DA1 for convenience) focuses on spoken and written text structure, and is

represented by subfields and approaches such as textlinguistics and Rhetorical Structure

Theory. The second tradition (let us call it DA2) focuses more widely on the relationship

between meaning, context and interaction, and is represented by such fields as pragmatics,

conversation analysis, interactional sociolinguistics and the ethnography of communication.

In the third tradition, the analytical focus broadens further still, towards the identification and

explication of systems of representation that, since the early work of Michel Foucault (1970,

1972), have been referred to across the humanities and social sciences as (countable noun)

5

discourses. It is this third tradition of discourse analysis (henceforth DA3) with which the

current paper is primarily concerned.

In this latter tradition, discourses are understood as social practices which

simultaneously function in two ways. Firstly, they provide humans as social beings with ways

of talking about things in the world that effectively ‘construct’ their knowledge and

understandings of these things. That is, they both enable and constrain our ability to talk or

think about an entity, process, concept or state of affairs in the world around us in a particular

way, or indeed to recognise something as having any ontological status at all (Teubert, 2010).

Secondly, by participating in a particular discourse, language users are not only viewing and

understanding (an aspect of) the world in a certain way, but also at the same time affiliating

themselves to, or claiming membership of, a particular social group (Gee, 2014). That is, in

taking up a particular discourse, I am effectively constructing myself (and will be constructed

by others) as having a particular professional or sociocultural identity, or as having particular

political or religious beliefs, or as occupying a particular stance in relation to an issue of

cultural or moral significance in my society, or as having certain aesthetic preferences, and so

on.

Crucially, both of these functions of discourse – the entity-constructing and the

identity-constructing – are predicated on the same two underlying assumptions: firstly, that

discourses must be characterized by symbolic forms that make them recognizable to

language users (and thus also to discourse analysts); and secondly, that at least some of these

symbolic forms must perforce to some degree be conventionalized, which is to say, repeated

in the same or similar surface form across multiple texts produced by different speakers,

writers and signers located in different places and at different points in time. It is the

identification of these conventionalized features of discourses, collectively referred to as

phraseologies (cf. Gledhill, 2000; Hunston, 2002), that forms the central task of the corpus-

based analysis of discourses as a distinctive research approach (see e.g. Orpin, 2005; Charles,

2006; Gabrielatos & Baker, 2008; Baker & McEnery, 2015; Potts et al., 2015; Brookes &

Harvey, 2016; Jaworska, 2016).

It is now widely recognized that corpus linguistics is ideally suited to the

phraseological analysis of discourses as described above. First of all, it is clearly preferable

for claims about discourses as large-scale social phenomena to be based on evidence from

corpora containing large numbers of texts produced by many different people than it is for

such claims to be based on readings of individual texts produced by single authors, no matter

how detailed, ingenious or persuasive these individual readings might be (Stubbs, 1997;

6

Toolan, 1997). Secondly, corpus linguistics provides researchers with a wide (and still

expanding) range of methods for identifying frequently occurring and/or statistically

significant language features in discourse-specific corpora (Baker, 2006; Taylor & Marchi,

2018; Kopaczyk & Tyrkkö, 2018). The only thing that this emergent field currently lacks at

present is any consensus regarding which theoretical models of language might be best suited

to the corpus-based analysis of discourses, or even whether such analyses need to be

grounded in any explicitly articulated theoretical model of language at all. The aim of the

remainder of this paper is to argue that some form of explicit theoretical grounding is indeed

desirable, and to make a case for CxG as a particularly apt choice for this purpose.

Before moving on, however, a brief note on nomenclature is in order. The reader may

already have noticed that I am using the term ‘corpus-based analysis of discourses’ to refer to

an approach that is more commonly known in the research literature either as ‘corpus-based

discourse analysis’, or ‘corpus-assisted discourse studies’. To clarify, I prefer to reserve the

label ‘corpus based discourse analysis’ for use as an umbrella term encompassing all three of

the main traditions of discourse analysis as discussed above. This is in recognition of the fact

that corpus-based methods have been applied to subfields as diverse as textlinguistics (e.g.

Hoey & O’Donnell, 2008, 2015), pragmatics (e.g. Aijmer, 2015, 2018; Rühlemann, 2018;

Rühlemann & Clancy, 2018), genre analysis (e.g. Biber et al., 2007; Upton & Cohen, 2009;

Groom & Grieve, 2019), narrative and conversation analysis (e.g. Toolan, 2009; Morton et

al., 2011; Koester, 2012; Rühlemann, 2014), as well as to the analysis of discourses in the

‘DA3’ sense described above. In this increasingly heterogeneous context, it is also useful to

have access to a more specific term as a means of referring explicitly and exclusively to

corpus-based research in the ‘DA3’ tradition, as opposed to other kinds of corpus-based

discourse research. For this purpose, I use the label ‘corpus-based analysis of discourses’

(henceforth CBADs) in preference to ‘corpus-assisted discourse studies’ (CADS), as it more

accurately describes the status of corpus analysis in this approach, at least as I conceive it.

Specifically, it signals a view of corpus analysis not as ‘assisting’ or ‘informing’ more

traditional approaches to ‘DA3’, but as the central activity of the whole enterprise.

4. Previous applications of CxG to CBADs

As mentioned briefly in the introduction to this paper, constructionist approaches have been

applied to a growing number of subfields within discourse analysis in recent years. In

7

particular, CxG frameworks have been successfully applied to studies in information

structure and discourse patterning (e.g. Lambrecht, 1994; Michaelis & Lambrecht, 1996;

Östman, 2005), context and pragmatics (e.g. Kay, 2004; Fried & Östman, 2005; Bergs &

Diewald, 2009; Jing-Schmidt, 2017), dialogue analysis and conversation analysis (e.g. Linell,

2009; Fischer, 2015) and register and genre studies (e.g. Antonopoulou & Nikiforidou, 2011;

Nikiforidou & Fischer, 2016; Hoffman & Bergs, 2018). However, all of these studies are

contributions to either the ‘DA1’ or ‘DA2’ traditions of discourse analysis as defined earlier.

‘DA3’, in contrast, has thus far been almost entirely neglected in CxG research. In fact, the

only published CxG study that addresses this third domain of discourse studies in any direct

way is a contrastive corpus-based analysis by Wulff et al. (2007) of the into-causative

construction (as in injury forced him into premature retirement or the Government was jolted

into action) in British and American English. Although Wulff et al. describe their study as a

comparison of national ‘varieties’ rather than ‘discourses’, it is reasonable to regard their

work as a ‘DA3’ study insofar as it pursues the idea that systematic differences in patterns of

construction usage may be indicative of deep-seated differences in British and American

worldviews and identities. As the authors themselves put it:

Bearing the basic assumption in mind that meaning construction is conceptualization … ,

which is, among other things, fundamentally shaped by a speaker’s cultural input, an

interesting question that has hitherto not been investigated in much detail in Construction

Grammar is: to what extent is the meaning potential of a construction variety-specific?

(Wulff et al., 2007: 278)

Using the methodology of collostruction analysis (Stefanowitsch & Gries, 2003; Gries &

Stefanowitsch, 2004), Wulff et al. (2007) study variations in the usage of the into-causative

construction across two newspaper corpora representing British and American English, and

find that this construction is significantly more frequently used in their British data to

construe causation in terms of (negatively evaluated) physical force or psychological

intimidation, whereas this construction is also frequently used in their American data (but not

in the British data) to construe causation in terms of (often positively evaluated) verbal

persuasion. They also find a marked preference for what they call ‘movement-initializing’

cause predicates (e.g. sting, provoke, chivvy) in British English, and a contrasting preference

for ‘movement-restricting’ cause predicates (e.g. snooker, rope, frustrate) in their American

corpus. The authors speculate that this latter contrast “may confirm the commonplace

8

perception that British culture lacks the strong and explicit emphasis on mobility as an

essential condition for a happy and free life as we find it in American culture” (Wulff et al.

2007: 279).

Wulff et al. are of course well aware of the limitations of basing such claims on the

analysis of individual linguistic features, and are thus careful to conclude by calling for

further studies “to see to what extent these differences also show up in other causative

constructions” (Wulff et al. 2007: 279). Regrettably, however, this challenge has not yet been

taken up by any other CxG researchers, nor have any subsequent studies systematically

pursued Wulff et al.’s more general hypothesis about the relationship between constructions

and cultural values.

This neglect of CBADs perspectives in the CxG research community is mirrored by a

corresponding lack of engagement with CxG on the part of CBADs researchers. Indeed, to

my knowledge the only CBADs study to have made explicit use of a constructionist

analytical framework thus far is an unpublished PhD thesis by Turo Hiltunen (2010). As with

Wulff et al. (2007), the aim of Hiltunen’s study is to investigate whether differences in

construction usage across corpora representing different cultural groups can be interpreted as

reflecting differences in the ways in which these cultural groups construe (aspects of) the

world around them. However, whereas Wulff et al. (2007) draw broad comparisons between

speakers of American and British English, Hiltunen’s focus is on the much more specialized

discourses of different academic disciplines, as represented by four corpora of research

articles in the fields of Law, Literary Criticism, Medicine and Physics respectively.

The part of Hiltunen’s thesis that makes the most thoroughgoing use of a CxG

approach centres on a comparative analysis of the as-predicative construction (as in The

Egyptians regarded time as a succession of recurring phases or The fire is being treated as

arson; both from the BNC) across the abovementioned four corpora. Hiltunen finds

significant differences in the usage of this construction across the four fields in question, and

argues that these differences can be mapped onto canonical distinctions (originating in the

higher education studies literature) between the epistemologically ‘hard’ (i.e. positivist)

academic fields of the natural and physical sciences and the epistemologically ‘soft’ (i.e.

hermeneutic) fields of the humanities and social sciences (Biglan, 1973a, 1973b).

Researchers in ‘hard’ fields tend to work within universally agreed and sharply defined

research paradigms, and regard the business of knowledge creation in their discipline as a

matter of discovering facts by conducting empirical (and usually experimental) research. In

Becher & Trowler’s (2001) highly influential terms, the epistemologies of ‘hard’ fields can

9

thus be described as “cumulative”, “atomistic” and “concerned with universals, quantities,

[and] simplification”. This being the case, Hiltunen argues, it is not surprising to find that the

as-predicative is predominantly used by physicists and medical researchers “for reporting

their own research activities” (p.238), as illustrated in Examples 1 and 2 (taken from

Hiltunen’s data) below:

(1) Individual irradiated cells were classified as either clonogenic or nonclonogenic

(Physics)

(2) We used the midpoint of LVAD enrollment as the dividing point for comparing the 2

cohorts. (Medicine)

In ‘Soft’ fields, in contrast, there is no single agreed theoretical paradigm for researchers to

conform to, and knowledge creation is seen not as a matter of discovering facts or solving

problems, but of offering or evaluating different perspectives on perennial questions or

canonical texts. In Becher & Trowler’s (2001: 36) terms, the epistemologies of ‘soft’ fields

can thus be described as “[r]eiterative”, “holistic” and “concerned with particulars, qualities

[and] complication”. These qualities, according to Hiltunen (2010), explain why the as-

predicative “is much more likely to be used for either advancing a particular claim, or

reporting an assertion that has previously been made by someone else”, as illustrated in

Examples 3 and 4:

(3) Casebooks generally treat the economic approach as an “exotic perspective” (Law)

(4) We may read it, I propose, either as allegory about the way in which the intensities of

experience felt as deeply private are also a social gesture, or as aesthetic allegory

about another kind of publication of private vision. (Literary Criticism)

As Hiltunen (2010) also notes, the underlying semantics of this construction (crudely,

‘somebody conceptualizes something as something’) make it a particularly apt linguistic

resource for expressing statements of interpretation or evaluation. This in turn explains why

the as-predicative construction occurs by far the most frequently in Hiltunen’s corpus of

academic writing in Literary Criticism, a discipline whose central concern is with the

interpretation and evaluation of literary texts.

10

In summary, the review of previous research above indicates that constructionist

approaches have already been successfully applied in empirical studies of discourses at very

different levels of socio-cultural specificity. However, it must be conceded that neither of the

studies previous discussed above is a full synthesis of constructionist and discourse analytic

perspectives. Wulff et al. (2007) make no direct reference to discourse analytic concepts in

their study, and Hiltunen (2010) imports aspects of CxG into his discourse research without

fully committing to CxG as a theory; indeed, he explicitly describes his overall approach as

“theory-neutral” (Hiltunen 2010: 91). In the following section, therefore, it will be necessary

to offer a more detailed case study analysis, not only as a means of demonstrating more

concretely how CxG can be applied in a CBADs context, but also as an inaugural

contribution to what I hope will be a new tradition of research.

5. Case study: epistemological variation and the WAY IN WHICH construction

The case study example to be presented in this section focuses on the analysis of academic

discourses. Following essentially the same line of enquiry as Hiltunen (2010), it tests the

hypothesis that systematic patterns of variability in construction usage will be observed

across corpora representing different academic discourses, and that these patterns of

variability can be interpreted as both reflecting and constructing fundamental epistemological

differences across these discourses.

The construction chosen for analysis here will be referred to in this paper as the WAY IN

WHICH construction. This construction is illustrated by Examples 5 – 7 below, all of which

have been taken from the BNC:

(5) she will have to promote changes in the way in which the markets go about their

business.

(6) The ways in which meaning is expressed must always be central to any language

study

(7) … the need, for example, to reform the appallingly undemocratic way in which

Northern Ireland business is handled at Westminster

11

This construction has been chosen for present purposes because it has already been identified

(albeit in non-constructionist terms) as a characteristic feature of written academic English,

notably by Biber et al. (1999). Multi-word sequences based on the trigrams way in which and

ways in which have also been identified as frequently occurring in several studies of

academic discourses using a ‘lexical bundles’ methodology (e.g. Biber et al., 2004; Cortes,

2004; Hyland, 2008; Durrant, 2017). However, as this construction has never been previously

discussed or even identified as a construction in the previous CxG literature, it will first be

necessary to establish that it has constructional status in its own right, and is not merely a

‘construct’ of a more general construction located at a higher level of schematicity.

5.1 A closer look at the WAY IN WHICH construction

On the face of it, the morphosyntactic pattern underlying Examples 5 – 7 above would seem

to constitute a clear form-meaning pairing, and thus a strong candidate for

‘constructionhood’. In terms of its form, this pattern is a complex noun phrase consisting of

three main elements: the head noun WAY, followed by the relative adverbial in which,

followed by a finite clause. The head noun itself is open to all the standard options pertaining

to countable nouns, i.e. it may occur in singular or plural form and may be preceded by a

definite or indefinite article, or (when plural) by no article at all. The head noun is also often

premodified adjectivally, as in Example (7) above; in the BNC, for instance, the node term

way/ways in which is preceded by at least one adjective in nearly a quarter of all instances.1

In terms of its meaning and function, this candidate construction serves to focus attention on

how the process expressed in the finite clause happens or is the case. If the head noun is

premodified (as in Example (7) above), it is the premodifying element(s) that provide the

‘manner’ information. If the head noun is not premodified (as in Examples (5) and (6)

above), the ‘manner’ meaning is usually left entirely unspecified in the construction itself,

and is thus either assumed as already shared knowledge, or predicted (cf. Tadros, 1994) as

information that will be provided at some subsequent point in the unfolding text. In some

cases, however, the ‘manner’ meaning is stated explicitly in the main clause element of the

construction itself. To illustrate this, consider Example (8) below:

(8) The debate was characterized by widespread dissatisfaction with the way in which the

UK had announced Sir David's resignation without appointing a replacement (BNC)

12

As can be seen, the clause-final prepositional phrase ‘without appointing a replacement’

describes how Sir David’s resignation was announced, and in so doing also explains why

there was ‘widespread dissatisfaction’ with the way this resignation was announced.

Taken together, these syntactic and semantic properties can be represented

schematically, as shown in Figure 1.

Semantics:

MANNER PROCESS

Syntax:

[DET] [ADJ] WAY in which Finite Clause

Examples: the way in which

the charges are levied

The ways in which ideas and beliefs in societies constrain the behaviour of their members

the appallingly undemocratic way in which

Northern Ireland business is handled at Westminster

the way in which the UK had announced Sir David's resignation without appointing a replacement

Figure 1: Schematic representation of the WAY IN WHICH construction

One significant shortcoming of the above account, however, is that it fails to acknowledge

that ‘WAY in which + finite clause’ is just one of three variant ‘manner head noun’ forms

featuring WAY as head, the other two being ‘WAY + that + finite clause’ and ‘WAY + finite

clause’ (i.e. with no relativizing element). Consider, by way of illustration, Examples 9 – 11

below:

(9) she will have to promote changes in the way in which the markets go about their

business.

(10) she will have to promote changes in the way that the markets go about their business.

(11) she will have to promote changes in the way the markets go about their business.

13

These three variants are syntactically very similar, and there seems to be no discernible

change in semantics or in level of formality (in register terms) from one variant to the next.

Indeed, it could even be argued that these three variants are better understood as

allostructions, “variant structural realizations of a construction that is left partially

underspecified” (Cappelle, 2006: 18; cf. Perek, 2015) – the MANNER HEAD NOUN

construction, perhaps. What, then, is the justification for focusing exclusively on the ‘WAY in

which + finite clause’ variant as a construction in its own right, as I propose to do in the case

study analysis below?

The simplest answer to this question is that ‘WAY in which + finite clause’ can be

viewed as satisfying the last of Goldberg’s three criteria as discussed earlier: “Any linguistic

pattern is recognized as a construction … as long as [it] occur[s] with sufficient frequency”

(Goldberg, 2006: 5; emphasis added).

That is, ‘WAY in which + finite clause’ qualifies as a construction because its

frequency profile – or, more precisely, its frequency distribution profile – indicates that

English speakers perceive it as having a distinctive status, which in turn means (from a

usage-based perspective, at least) that it forms a part of speaker knowledge that must be

accounted for on its own terms.

It is important to note that this interpretation goes somewhat beyond Goldberg’s

(2006) formulation as quoted above. Goldberg’s concept of “sufficient frequency” is usually

interpreted in terms of absolute frequency in the language as a whole, and Goldberg herself

makes no explicit claim that a strong association between a pattern and a particular genre in

itself constitutes evidence for constructionhood in that genre. It is nevertheless my contention

that the criterion of sufficient frequency can plausibly and usefully be interpreted from a

usage-based perspective as meaning ‘occurring with sufficient frequency in a particular

discourse or genre’.

Empirical support for this claim comes from Biber et al. (1999: 628), who present a

comparative corpus-based analysis of the three variant forms ‘WAY in which + finite clause’,

‘WAY + that + finite clause’ and ‘WAY + finite clause’. While Biber et al (1999) note that all

three of these variant forms are frequently attested in academic prose, they find that ‘WAY in

which + finite clause’ – but not the other two variant forms – is found to be exclusively

associated with academic prose as a broad register variety. To check this finding, I also

carried out a distributional analysis for the search query way in which in the BNC, using the

‘Distribution’ function in BNCweb. The relevant section of this analysis is provided in

Appendix 1. Although the BNC analysis paints a somewhat less extreme picture, Biber et al.

14

(1999)’s essential observation that the WAY IN WHICH construction is strongly associated

with academic writing (and dispreferred in news journalism, prose fiction and conversation)

holds true. This suggests that it seems to be part of English speakers’ knowledge that ‘WAY in

which + finite clause’ is strongly primed for use in academic prose, and equally strongly

negatively primed for use in these other registers (Hoey, 2005). On the basis of these

observations, therefore, I propose that it is valid to regard the morphosyntactic pattern ‘WAY

in which + finite clause’ as a construction, and to focus exclusively on this construction in the

analysis that follows.

5.2 Data and methods

The question to be investigated here follows up on the LGSWE account of the WAY IN

WHICH construction by asking whether its usage is quantitatively and qualitatively variable

across different academic disciplines, and if so, whether such variability can be interpreted as

both reflecting and helping to construct different ‘ways of knowing’ in different fields of

study. The methodology of the study is based on concordance analyses of the node terms way

in which and ways in which across six corpora of academic research articles representing six

very different disciplinary discourses: Cell Biology and Electrical Engineering, representing

the physical and natural sciences; Economics and Sociology, representing social sciences;

and English Literature and History, representing the humanities. A summary of the

composition of each corpus is provided in Appendix 2.

The two humanities corpora were compiled by the author, and consist of articles

published in leading journals between 2010 and 2014. The other four corpora also contain

articles from leading journals, but were compiled by another researcher (Oakey, 2008,

2009)2, and represent a somewhat older sampling period (from 1998 to 2004). It should also

be acknowledged here that the six corpora are of varying size and composition in terms of the

number of texts and the number of word tokens they contain, as well as in the number of

journals sampled for each discipline. These differences notwithstanding, I would submit that

the six corpora remain adequately comparable for the purposes of the current paper.

An initial quantitative analysis was carried out in order to establish the frequency

distribution of the WAY IN WHICH construction across the six corpora. Singular and plural

forms of the construction were analysed separately. A qualitative concordance analysis of the

full set of hits for each query in each corpus was then conducted, following a coding

15

procedure that will be explained in more detail below. All corpus analyses were carried out

using AntConc version 3.5.0 (Anthony, 2018).

5.3 Quantitative analysis

The results of the initial frequency distribution analysis of the WAY IN WHICH construction

across the six corpora are presented in Table 1. Also provided in Table 1 is a set of

benchmark figures for general reference purposes. These figures (representing ‘academic

prose’ as a broad register category along the lines of the LGWSE analysis discussed earlier)

were obtained by adding the results for all six corpora together and dividing the totals by six.

Table 1: Frequency distribution of the WAY IN WHICH construction across six discipline-specific

corpora Corpus Frequency per million words

way in which ways in which TOTAL Sociology 56.98 114.75 171.73 English Literature 50.97 69.09 120.06 History 34.49 44.74 79.23 ‘Academic prose’ 26.86 39.70 66.57 Economics 15.36 8.08 23.44 Cell Biology 1.89 1.11 3.00 Electrical Engineering 1.49 0.44 1.93

Far from being characteristic of ‘academic prose’ in general, the current analysis finds that

the WAY IN WHICH construction varies dramatically and systematically across the six

disciplinary discourses studied here. The most striking general observation is that the WAY

IN WHICH construction occurs very frequently in the humanities, and even more so in the

‘soft’ social science field of sociology, but is actually uncharacteristic of academic discourse

in hard knowledge fields.

Although the figures for History are only slightly higher than the benchmark

frequency, the WAY IN WHICH construction occurs at almost twice the average rate in the

discourse of English Literature, and is nearly three times more frequent than the benchmark

figure in Sociology. In stark contrast to this, the WAY IN WHICH construction occurs in the

discourse of Economics only a third as frequently as might be expected from a general

16

register perspective, and – perhaps most strikingly of all – it is hardly ever attested in pure

science discourses such as Cell Biology and Electrical Engineering.

These patterns of variability are highly consistent with Becher & Trowler’s (2001)

characterisations of epistemological variation across academic disciplines as discussed earlier

in this article. Given that the principal function of the WAY IN WHICH construction is to

focus attention on statements of manner, that is, on how things manifest themselves, it is not

surprising to find that this construction is strongly preferred in ‘soft’ disciplines, where the

creation of new knowledge tends to focus on “particulars, qualities [and] complication.”

Becher & Trowler’s account also offers to explain why the WAY IN WHICH construction is

hardly ever used in ‘hard’ disciplines. In these discourses, the focus is much more on

“universals, quantities, [and] simplification”; that is, researchers are principally interested in

discovering new facts about the external world, and not so concerned with nuancing

understandings of already established phenomena.

This emphasis in humanities fields on providing detailed, nuanced and complex

accounts of disciplinary phenomena (and the corresponding drive towards simplification and

generalization in the sciences) also seems to offer a plausible explanation for the other main

observation emerging from the analysis presented in Table 1: that the plural variant ways in

which occurs more frequently than does the singular variant way in which in Sociology,

English Literature and History, whereas the reverse is the case in Economics, Cell Biology

and Electrical Engineering.

5.4 Qualitative analysis

The quantitative analysis reported above does not tell whole story, however. A qualitative

inspection of the concordance data reveals a more subtle, but no less systematic, difference in

the usage of the WAY IN WHICH construction across the hard/soft epistemological divide.

As explained in Section 5.1, the canonical function of the WAY IN WHICH

construction is to comment or focus attention on how a phenomenon occurs or is the case.

That this phenomenon does occur or is the case is, in general, not at issue. To illustrate,

consider Examples 12 - 14:

(12) the way in which Cabinet minutes were recorded, Attlee claimed, stressed the role of

the prime minister (History)

17

(13) the specific organization of welfare state institutions and the way in which policy

measures are implemented also affect the structuring of the life course. (Sociology)

(14) Caleb Williams's original publication context was thus crucial to the ways in which

the novel was received. (English Literature)

In Examples (12) and (13), it would be decidedly odd to deny that senior government

meetings are minuted, or that government policies are (at least sometimes) implemented.

Likewise, the novel in question in Example (14) is known to have caused a political stir on its

publication in 1794, so there can be no question that it was ‘received’ in a variety of ways by

the reading public of its day.

However, closer analysis of individual concordance lines reveals that there are

instances where the statement in the main clause does not have this ‘to be taken for granted’

status. Consider Examples 15 – 17:

(15) Work of the last decade has explored the ways in which Mrs. Dalloway’s prose style

constitutes a critique of liberal imperialism’s dominant epistemology. (English

Literature)

(16) The idea that the Kemalist project of modernization as the bearer of democracy,

rights, justice, and truth can provide the vehicle for democratization is called into

question by the way in which it positions minorities and rights-asserting groups as

particularisms subordinate to its own universalism. (Sociology)

(17) Whereas Eric Evans and many subsequent historians have viewed tithes as illustrative

of the way in which the clergy were increasingly distanced and separated from lay

society, this article has demonstrated that … tithe relationships continued to bind the

inhabitants of every parish. (History)

In Example (15), the reader does not have to be an expert on Virginia Woolf’s fiction (or

even need to know that Mrs Dalloway is a novel by Virginia Woolf) to recognise that not

every literary critic will agree with the claim that the prose style of this novel constitutes a

critique of liberal imperialism’s dominant epistemology. Similarly, there is a clear

18

discrepancy in Example (16) between the Kemalist project’s decidedly upbeat view of itself

and the more critical view of this movement that is being put forward by the writer. The non-

factive status of the underlined proposition in Example (17) is even more obvious; the writer

here explicitly states that their own research undermines the arguments previously proposed

by Eric Evans and other historians.

In these examples, then, the main purpose of the WAY IN WHICH construction

seems to be the opposite of the canonical usage as described earlier; that is, the writer seems

to be using the construction not to focus the reader’s attention on how things manifest

themselves, but to argue that these things are actually the case at all. Accordingly, we can call

this usage proposition-focused, in contrast to the canonical manner-focused variant.

Interestingly, it transpires that proposition-focused instances can be identified by

means of a simple substitution test. In all cases, the three word sequence way in which can be

substituted by the two-word sequences fact that or idea that, without changing the overall

meaning of the text, as illustrated in Examples (18)–(20).

(18) Eric Evans and many subsequent historians have viewed tithes as illustrative of the

fact that the clergy were increasingly distanced and separated from lay society …

(19) Work of the last decade has explored the idea that Mrs. Dalloway’s prose style

constitutes a critique of liberal imperialism’s dominant epistemology.

(20) The idea that the Kemalist project of modernization as the bearer of democracy,

rights, justice, and truth can provide the vehicle for democratization is called into

question by the fact that it positions minorities and rights-asserting groups as

particularisms subordinate to its own universalism.

This substitution does not work with canonical ‘manner-focused’ instances of the WAY IN

WHICH construction; in such cases, the reworded statement either makes no sense at all, or

entirely changes the original meaning, as can be seen in Examples (21) and (22) below:

(21) *The fact/idea that the axial-flux density changes radially across the active region

was calculated by 3-D FEA

19

(22) *the fact/idea that Cabinet minutes were recorded, Attlee claimed, stressed the role

of the prime minister

This substitution test provided the basis for a qualitative investigation into the distribution of

these two semantic variants across the six corpora. This investigation centred on an analysis

of full sets of concordance results for the search terms way in which and ways in which in

each corpus, and involved the manual inspection of the context of each individual

concordance line using AntConc’s File View tool. In lieu of an inter-rater reliability test, the

analysis adopted a stringently conservative policy of coding any instance that could not easily

be identified by the substitution test described above as manner-focused by default. The

results of this qualitative analysis are presented in Table 2.

Table 2: Distribution of manner-focused and proposition-focused WAY IN WHICH construction

variants across six discipline-specific corpora

Corpus Manner-focused Proposition-focused way in which ways in

which TOTAL

way in which

ways in which

TOTAL

English Literature

250/481 (52%)

240/651 (36.9%)

490/1132 (43.3%)

231/481 (48%)

411/651 (63.1%)

642/1132 (56.7%)

History 162/275 (58.9%)

194/358 (54.2%)

356/633 (56.2%)

113/275 (41.1%)

164/358 (45.8%)

277/633 (43.8.%)

Sociology 173/289 (59.9%)

433/582 (74.1%)

606/871 (69.6%)

116/289 (40.1%)

151/582 (25.9%)

267/871 (30.7%)

Economics 76/76 (100%)

40/40 (100%)

116/116 (100%)

0/76 (0%)

0/40 (0%)

0/116 (0%)

Cell Biology 29/29 (100%)

17/17 (100%)

46/46 (100%)

0/29 (0%)

0/17 (0%)

0/46 (0%)

Electrical Engineering

10/10 (100%)

3/3 (100%)

13/13 (100%)

0/10 (0%)

0/3 (0%)

0/13 (0%)

As can be seen, the proposition-focused usage is the exclusive preserve of the discourses of

“soft” disciplines; it is not attested in Economics, Cell Biology or Electrical Engineering at

all. Among the soft fields, this usage is particularly prominent in English Literature, where it

accounts for nearly 57% of all instances. The most likely explanation for this is that this

usage is particularly well suited to putting forward propositions that are purely interpretative,

and not claims to any kind of objective ‘truth’. The plausibility of interpretative claims in soft

fields in general, and in English Literature in particular, rests on the amassing of detailed

insights and observations drawn from close readings of texts. This being the case, it is

20

plausible to argue that the WAY IN WHICH construction is frequently used by literary critics

both to put forward an interpretative claim, and at the same time to signal that textual

evidence in support of this interpretation will be provided. That is, a statement such as

Example (23) simultaneously means both “I will argue that Capote eliminates, or at least

reduces, many of the strategic delays ordinarily imposed by the hermeneutic code” and “I

will provide examples to show how Capote eliminates, or at least reduces, many of the

strategic delays ordinarily imposed by the hermeneutic code”.

(23) I [will] then discuss the way in which Capote eliminates, or at least reduces, many of

the strategic delays ordinarily imposed by the hermeneutic code (English Literature)

To an unsympathetic observer, this rhetorical strategy may seem to be little more than a form

of argumentative sleight of hand – a way of smuggling in a dubious proposition under the

reader’s nose by disguising it as already shared knowledge. Within the context of humanities

discourses, however, this particular instance of construction usage is better understood (and

will no doubt be understood by members of these discourse communities) as a form of

epistemological shorthand (‘I will offer an interpretation and attempt to demonstrate its

plausibility through detailed analysis’), and not as an attempt to deceive.

In summary, the case study analysis presented in this section has shown that the WAY

IN WHICH construction is strongly associated with the disciplinary discourses of the

humanities and ‘soft’ social sciences, and can thus be regarded as a linguistic resource for

constructing and transmitting the epistemological values of these discourses. The analysis

also finds that, on closer qualitative inspection, this construction is exploited in more than

one way in the ‘soft’ fields. In some cases the underlying proposition is agreed or at least

uncontroversial, in which case the function of the construction is to introduce or predict the

addition of circumstantial details of manner. In other cases, it simultaneously serves to put

forward a proposition which is open to dispute, and to signal that evidence in support of this

proposition will be provided at a subsequent point in the text.

6. Remaining issues and objections

The aim of the empirical ‘proof of concept’ analysis presented in the previous section was to

demonstrate that usage-based construction grammar can be applied rigorously and

21

productively to the quantitative and qualitative corpus-based analysis of academic and other

specialized discourses. Irrespective of the persuasiveness of this case study analysis,

however, it is inevitable that misgivings will remain in some quarters about the more general

appropriacy of CxG as a theoretical framework for CBADs. Indeed, these misgivings may

well explain why constructionist approaches have not been taken up by more CBADs

researchers since the publication of the ground-breaking studies by Wulff et al. (2007) and

Hiltunen (2010) reviewed earlier. In this section, therefore, I wish to discuss two of the most

likely sets of reservations about the synthesis of discourse and constructionist research

perspectives advocated in this paper. Each of these will be discussed in turn below.

6.1 Terminological issues

Although a relatively new field of studies, CBADs is already overflowing with specialized

terminology (some of it at least partly overlapping) relating to the kinds of linguistic features

that researchers in this field typically focus on: collocations, collocation networks,

concgrams, formulaic sequences, grammar patterns, idioms, keywords, lexical bundles,

lexical items, lexical phrases, n-grams, p(hrase)-frames, phraseological units, semantic

sequences, skip-grams and units of meaning, among many others. While this embarrassment

of conceptual riches certainly attests to the vigour and eclecticism of CBADs as an emerging

field of research, it is also problematic in that it militates against the making of potentially

useful comparisons across empirical studies that use different sets of terms. From this

perspective, it is easy to understand why some CBADs researchers might be reluctant to

engage with CxG. To put it bluntly: the field is terminologically overburdened already, so

why complicate matters even further by importing yet another battery of technical terms for

what appear to be essentially the same underlying phenomena?

In response to this, I would like to suggest that the reverse is actually the case: that

adopting a constructionist approach to CBADs actually simplifies the conceptual landscape,

and clarifies the relationships among concepts at different levels of analysis. As a first step in

this argument, it is necessary to establish that the objection sketched out above rests on a

fundamentally misguided assumption: specifically, that there is a broad theoretical

equivalence between the corpus-based terms and concepts listed in the previous paragraph

(e.g. collocations, grammar patterns, keywords, lexical bundles, p-frames, units of meaning,

etc.) on the one hand, and the conceptual apparatus proposed by CxG on the other. This is

22

very far from the case in reality. On the contrary, whereas the term ‘construction’ is the

central concept in a full and coherent theory of language as a symbolic system of usage-based

form-meaning pairs, none of the terms listed in the previous paragraph has any formal

theoretical status whatsoever; they are all descriptive terms, firmly rooted in the empirical

business of corpus analysis. In some cases, of course, this is nothing more than a statement of

the obvious. Keywords, lexical bundles, n-grams and p-frames, for example, are clearly not

theoretical linguistic concepts in any strict sense at all; rather, they are simply the names of

automatic extraction procedures that do double duty as descriptive labels for the kinds of

usage data yielded by these procedures. However, this argument also holds true for the less

obviously ‘methodological’ of the corpus-based concepts listed above. To take just one

example, ‘(extended) units of meaning’ (Sinclair, 1996, 2004) is one of the most important

and influential concepts ever to have emerged from the field of corpus linguistics, but it is a

concept that has never been formally defined, nor has it ever been placed within a

comprehensive theory of language. What is clear from the examples presented in Sinclair’s

own work (e.g. his oft-cited analyses of naked eye, true feelings, brook and place) is that

units of meaning are idiomatic multi-word phrases which are both highly patterned and yet

open to substantial internal variation, and whose meanings are a property of the extended unit

as a whole and not reducible to any of its constituent parts. Sinclair (1991, 1996, 2004) points

out that such units are very frequent (and probably the default) in natural language data, but

does not say how far the concept of units of meaning applies to language as an overarching

system. This leaves a range of fundamental questions unanswered. If the notion of units of

meaning does not in itself constitute a full and comprehensive account of language, then what

are its limits? What other kinds of linguistic unit co-exist with units of meaning, and how (if

at all) are they interrelated?

In summary, then, corpus linguistic concepts such as units of meaning, collocations,

keywords and lexical bundles describe fundamentally important observations about language

in use obtained from corpora, but it must be recognized that they do not – either singly or in

combination – amount to anything like a formal or comprehensive theory of language of the

kind proposed by CxG. They are ways of talking about what language does, and not ways of

talking about what language is. This should not be seen as a problem, however. On the

contrary, my argument is that recognising a distinction between constructions as theoretical

and explanatory and corpus-based concepts as empirical and descriptive is enormously

helpful and liberating. Not only does it provide clearer distinctions between objects of

analysis at different levels of analysis, but it also provides a coherent means of articulating

23

relationships among these. As mentioned above, constructions are abstract theoretical

generalisations, and thus cannot be observed directly in corpora. Conversely, corpus-derived

features such as collocations, lexical bundles and units of meaning are empirical observations

which lack any explicit theoretical status. Each thus dovetails with the other, yielding a two-

level analytical model in which a variety of qualitative and quantitative observational

methods are used to extract empirical usage data from corpora, which are then reinterpreted

theoretically as constructions. This relationship is represented diagrammatically in Figure 2.

LEVEL OF ANALYSIS OBJECT OF ANALYSIS Language features Constructions (theoretical) Language usage Quantitative observations Qualitative observations (empirical) (e.g. keywords, and/or (e.g. units of meaning, lexical bundles, grammar patterns, p-frames …) semantic sequences …)

Figure 2: Levels and objects of analysis in constructionist CBADs

A remaining objection at this point might be that CBADs already has a well-established

general term that covers all the features identified by corpus analysis: the concept of

phraseology, influentially defined by Hunston (2002: 137) as “the tendency of words to

occur, not randomly, or even in accordance with grammatical rules only, but in preferred

sequences.” (cf. Cowie, 1998; Granger & Paquot, 2008; Gries, 2008). While it is certainly the

case that phraseology is the most popular general term for referring to the characteristic

linguistic features of discourses, I would argue that there are two fundamental problems with

this concept as it is currently used in CBADs, both of which can be solved by replacing it

with the concept of constructions.

Firstly, and to recapitulate an earlier argument, ‘phraseology’ turns out on close

inspection to be yet another descriptive term, and not a fully-fledged theoretical concept on a

par with the concept of ‘construction’. It tells us something important about what language

does (i.e. it tends to occur in preferred sequences), but it does not constitute a formal claim

24

about what language is. As Hunston (2002: 147) puts it, although “phraseology is extremely

pervasive … phraseology alone cannot account for how sentences or utterances are made

up.” The concept of construction, in contrast, does encompass and account for all kinds of

language features, from individual words to complete sentences and utterances, and from

fully lexicalized multi-word units to entirely schematic grammatical patterns. In so doing, it

renders questions about the boundary between the phraseological and the non-phraseological

in language moot, and thus effectively renders the concept of phraseology redundant. (On this

point, it is telling that ‘phraseology’ does not even warrant an index entry in Hoffman &

Trousdale (2013), the standard reference work on CxG).

Secondly, a review of the CBADs literature finds that phraseology is also frequently

used to refer in a much looser and more general way to the whole set of conventionalized

language features that have to be learned by anyone who wishes to participate legitimately in

any given discourse. This ‘specialized literacy practice’ meaning of the term, defined by

Gledhill (2000: 1, 202) as “the preferred way of saying things in a particular discourse” is

strongly conterminous with the lay meaning of the term as found in standard dictionary

definitions.3 This conceptualization of phraseology is very useful for CBADs, not only

because it captures the idea that discourses have recognisable linguistic characteristics, but

also because it recognises that not all contributions to a discourse will be formulaic or even

frequently repeated within that discourse (Teubert, 2010). Felicitously, replacing the lexico-

grammatical concept of phraseology with the concept of construction means that, instead of

being abolished altogether, phraseology can be repositioned at this higher level of theoretical

abstraction. This in turn allows us to expand the analytical model proposed in Figure 3 into a

full and coherent set of terms and concepts for constructionist approaches to CBADs. This

conceptual model is presented in Figure 3.

25

LEVEL OF ANALYSIS OBJECT OF ANALYSIS Social practices Discourses Literacy practices Phraseologies Language features Constructions (theoretical) Language usage Quantitative observations Qualitative observations (empirical) (e.g. keywords, and/or (e.g. units of meaning, lexical bundles, grammar patterns, p-frames …) semantic sequences …) Figure 3: A conceptual model for constructionist approaches to CBADs

Before moving on, it should be noted that the idea of repositioning the concept of

phraseology as presented here is not an entirely new one. On the contrary, very similar

proposals can be found in Gledhill (2000: 203; see also Gledhill, 2011), and the model

presented in Figure 3 is directly inspired by Gledhill’s example. The main differences

between the two models are that Gledhill’s is based around the Hallidayan concept of

“lexicogrammatical patterns” rather than constructions, and does not posit an explicit

theoretical distinction between phraseology and discourse of the kind envisaged here.

6.2 Objections to CxG’s ‘cognitive commitment’

The second set of objections that I wish to discuss here centres on the CxG’s close and

abiding relationship with cognitive linguistics. As acknowledged at the beginning of this

article, CxG originated as a theoretical alternative to Chomskyan generativism, and shares the

generativist aspiration to understand how language is learned by, represented in, and accessed

from, the mind of the language user. At issue here, then, is the question of whether and to

26

what extent this ‘cognitive commitment’ makes CxG fundamentally incompatible with

CBADs as a socially-oriented tradition of research.

For some scholars, the answer to this question is a resounding ‘yes’. Teubert (2005:

8), for example, argues not only that “corpus linguistics and cognitive linguistics are two

complementary, but ultimately irreconcilable paradigms”, but also that

[t]he mind is no more an object of scientific investigation than the soul … [and] any attempt

to describe what goes on in the mind has to fail because it will run into contradictions that

cannot be resolved.

(Teubert, 2010: 43)

For others, objections to CxG’s cognitivist philosophical underpinnings may be motivated by

pragmatic rather than ideological concerns. Irrespective of their views on the cogency or

otherwise of cognitive theory, researchers who are principally interested in studying how

people use language to achieve particular social goals and purposes may understandably see

little point or benefit in engaging with a theoretical approach that is avowedly committed to

understanding language as a system of knowledge in the minds of individual language users.

The simplest response to both of these objections is to point out that not all

construction grammarians regard the ‘cognitive commitment’ as obligatory, or even

necessarily relevant in some fields of linguistic enquiry. Here, for example, is Hilpert (2018:

24-25):

While the notion of a schematic pattern strongly invites the idea of a speaker who mentally

represents an abstraction over many concrete usage events, it is actually possible for an

analyst to remain agnostic about this issue. It can be left open whether abstract patterns have

psychological reality or whether they are in fact merely posited as theoretical constructs,

much in the way that linguistic notions such as ‘head’, ‘clitic’, or ‘subordinating conjunction’

serve as useful labels for phenomena that capture insights about a linguistic system but that

may not correspond to psychologically real categories in the minds of speakers.

Hilpert’s point is made in the context of a review of constructionist approaches to the study

of language change, but it applies equally well to the field of discourse analysis. Indeed, the

agnostic stance sanctioned by Hilpert is essentially the same as that already adopted by

Hiltunen (2010) in his “theory-neutral” study of constructional variability in academic

27

disciplinary discourses. The position taken by this paper is similar, but not identical to

Hiltunen’s. Specifically, it does endorse the CxG account of language acquisition, storage

and processing, but leaves such matters entirely implicit, on the grounds that they are not

directly germane to the research questions that were pursued in Section 5 of this paper.

7. Conclusion

In this paper I have advocated usage-based Construction Grammar as a powerful, flexible and

productive framework for carrying out corpus-based research on discourses, defined as ways

of saying that are also ways of knowing, being and acting in the social world. The practical

case study example at the heart of the paper applied precisely such an approach in an analysis

of a newly identified construction, the WAY IN WHICH construction, across six corpora

representing a range of different academic disciplinary discourses. The analysis found that

this construction is very frequently used in humanities and ‘soft’ social sciences discourses,

but hardly used at all in the ‘hard’ sciences and quantitative social sciences. The analysis also

identified two distinct usage variants of the WAY IN WHICH construction –manner-focused

and proposition-focused – and found that the latter is exclusively associated with the

interpretative epistemologies of ‘soft’ disciplines. Taken together, these observations provide

empirical support for the theoretical claim that the epistemologies and phraseologies of

academic disciplines are mutually constitutive.

As well as demonstrating the practical viability of applying CxG to CBADs, I have

also offered a number of more general theoretical arguments in favour of adopting a

constructionist approach to the analysis of discourses. In particular, I have argued that, far

from adding unnecessary complexity into the field, basing CBADs around the central

theoretical concept of ‘construction’ allows us to clarify and simplify a number of key terms

and concepts, and to bring them into a more coherent overall relationship. Finally, following

Hilpert (2018), I have argued that it is perfectly possible for CBADs researchers to embrace

usage-based CxG as an overarching theoretical framework without needing to affiliate to, or

at least engage explicitly with, its much-vaunted ‘cognitive commitment’.

While this paper has generally focused throughout on showing how CBADs can

benefit from CxG, I would also like to propose that the reverse is also the case; that is, that

CxG stands to benefit substantially from a fuller engagement with the wider perspectives

revealed by CBADs. One of the implications of the case study analysis presented in this

28

paper, for example, is that it indicates that Goldberg’s (2006) “sufficient frequency” criterion

would benefit from being expanded so that it recognises the observation of a skewed

distribution of a feature across discourses as an indicator of constructionhood, as well as (or

perhaps even instead of) the frequency of a construction in absolute terms.

Of course, it would be unrealistic as well as presumptuous to argue that all corpus-

based discourse analysts should become construction grammarians, or that all construction

grammarians should cultivate a keen interest in the analysis of discourses, and it has not been

my intention in this paper to propose this. What I am claiming, however, is that CxG is far

more compatible with CBADs than it might at first glance seem, and that there are in fact

good reasons for regarding CxG as offering the most elegant and compelling theoretical

framework for CBADs currently available. If these arguments encourage at least some

discourse researchers to consider engaging more fully with CxG than they may have done

hitherto, then this paper will have served its main purpose.

Acknowledgements

I wish to thank the three anonymous reviewers of this paper for their helpful comments and

suggestions.

Notes

1. The search query (way | ways) in which returns 5279 hits for the whole BNC. Searching for _{A}

(way | ways) in which (i.e. the same query fronted by an adjective) returns 1274 hits, i.e. 24.13% of

all instances.

2. I am grateful to David Oakey for generously allowing me to use his data for this study.

3. The Oxford Online Dictionary, for example, defines phraseology as “A particular mode of

expression, especially one characteristic of a particular speaker or subject area. ‘legal phraseology’.”

https://en.oxforddictionaries.com/definition/phraseology

29

References

Aijmer, K. (2015). General extenders in learner language. In N. Groom, M. Charles & S.

John (Eds.), Corpora, Grammar and Discourse: In honour of Susan Hunston (pp.

211-234). Amsterdam: John Benjamins.

Aijmer, K. (2016). Pragmatic markers as constructions. The case of anyway. In G.

Kaltenböck, E. Keizer & A. Lohmann (Eds.), Outside the Clause: Form and function

of extra-clausal constituents. [Studies in Language Companion Series 178] (pp. 29-

58). Amsterdam: John Benjamins.

Aijmer, K. (2018). ‘Corpus pragmatics: From form to function.’ In A.H. Jucker, K.P.

Schneider & W. Bublitz (Eds.), Methods in Pragmatics (pp. 555-586). Berlin: De

Gruyter Mouton.

Anthony, L. (2018). AntConc (Version 3.5.0 for Mac OSX). [Computer Software].

http://www.laurenceanthony.net/software/antconc/

Antonopoulou, E., & Nikiforidou, K. (2011). Construction grammar and conventional

discourse: A construction-based approach to discoursal incongruity. Journal of

Pragmatics, 43(10), 2594-2609.

Baker, P. (2006). Using Corpora in Discourse Analysis. London: Continuum.

Baker, P., & McEnery, T. (Eds.) (2015). Corpora and Discourse Studies: Integrating

Discourse and Corpora. Basingstoke: Palgrave Macmillan.

Becher, T., & Trowler, P. (2001). Academic Tribes and Territories: Intellectual Enquiry and

the Culture of Disciplines. (2nd ed.). Buckingham: The Society for Research into

Higher Education and Open University Press.

Bergs, A., & Diewald, G. (Eds.) (2009). Contexts and Constructions. Amsterdam: John

Benjamins.

Biber, D., Connor, U., & Upton, T.A. (2007). Discourse on the Move: Using Corpus Analysis

to Describe Discourse Structure. Amsterdam: John Benjamins.

Biber, D., Conrad, S., & Cortes. V. (2004). If you look at . . . : Lexical bundles in university

teaching and textbooks.’ Applied Linguistics, 25, 371-405.

Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). The Longman

Grammar of Spoken and Written English. Harlow: Pearson Education.

Biglan, A. (1973a). The characteristics of subject matter in different scientific areas. Journal

of Applied Psychology, 57(3), 195-203.

30

Biglan, A. (1973b). Relationships between subject matter characteristics and the structure and

output of university departments. Journal of Applied Psychology, 57(3), 204-213.

Boas, H.C. (2013). Cognitive Construction Grammar. In T. Hoffman & G. Trousdale (Eds.)

The Oxford Handbook of Construction Grammar (pp. 233-252). Oxford: Oxford

University Press.

Boas, H.C., & Sag, I. (Eds.) (2012). Sign-Based Construction Grammar. Stanford: CSLI

Publications.

Brookes, G., & Harvey, K. (2016). Examining the discourse of mental illness in a corpus of

online advice-seeking messages. In L. Pickering, E. Friginal, & S. Staples (Eds.)

Talking at Work: Corpus-Based Explorations of Workplace Discourse (pp. 209-234).

Basingstoke: Palgrave Macmillan.

Cappelle, B. (2006). Particle placement and the case for 'allostructions'. Constructions.

https://www.constructions.uni-osnabrueck.de/wp-content/uploads/2014/06/2006-SI-

Cappelle22-80-1-PB.pdf

Charles, M. (2006). Phraseological patterns in reporting clauses used in citation: A corpus-

based study of theses in two disciplines. English for Specific Purposes, 25(3), 310-

331.

Cortes, V. (2004). Lexical bundles in published and student disciplinary writing: Examples

from history and biology. English for Specific Purposes, 23(4), 397-423.

Cowie, A.P. (1998). Phraseology: Theory, Analysis and Applications. Oxford: Clarendon

Press.

Croft, W., & Cruse, D.A. (2004). Cognitive Linguistics. Cambridge: Cambridge University

Press.

Durrant, P. (2017). Lexical bundles and disciplinary variation in university students’ writing:

Mapping the territories. Applied Linguistics, 38(2), 165-193.

Ellis, N. C. (2003). Constructions, chunking, and connectionism: The emergence of second

language structure. In C. J. Doughty & M. H. Long (Eds.), The Handbook of Second

Language Acquisition (pp. 63–103). Oxford: Blackwell.

Ellis, N.C., Römer, U., & O’Donnell, M.B. (2016). Usage-Based Approaches to Language

Acquisition and Processing: Cognitive and Corpus Investigations of Construction

Grammar. [Language Learning Monograph Series]. Malden, MA: Wiley.

Evans, V., & Green, M. (2006). Cognitive Linguistics: An Introduction. Edinburgh:

Edinburgh University Press.

31

Fillmore, C.J. (2013). ‘Berkeley Construction Grammar.’ In T. Hoffman & G. Trousdale

(Eds.), The Oxford Handbook of Construction Grammar (pp. 111-132). Oxford:

Oxford University Press.

Fillmore, C.J., & Kay, P. (1995) Construction Grammar. Manuscript, University of

California, Berkeley.

Fillmore, C.J., Kay, P., & O’Connor, M.C. (1988). Regularity and idiomaticity in

grammatical constructions: The case of let alone. Language, 64(3), 501-538.

Fischer, K. (2015). ‘Conversation, construction grammar, and cognition.’ Language and

Cognition, 7(4), 563-588.

Foucault, M. (1970). The Order of Things. London: Tavistock.

Foucault, M. (1972). The Archaeology of Knowledge. London: Tavistock.

Fried, M., & Östman, J.-O. (2005). Construction Grammar and spoken language: The case of

pragmatic particles. Journal of Pragmatics, 37(11), 1752-1778.

Gabrielatos, C., & Baker, P. (2008). Fleeing, sneaking, flooding: a corpus analysis of

discursive constructions of refugees and asylum seekers in the UK Press 1996-2005.

Journal of English Linguistics 36(1), 5-38.

Gee, J.P. (2014). An Introduction to Discourse Analysis: Theory and Method. (4th ed.).

Abingdon: Routledge.

Gilquin, G. (2015). The use of phrasal verbs by French-speaking EFL learners. A

constructional and collostructional corpus-based approach. Corpus Linguistics and

Linguistic Theory, 11(1), 51-88.

Gledhill, C. (2000). Collocations in Science Writing. Tubingen: Gunter Narr.

Gledhill, C. (2011). ‘The ‘lexicogrammar’ approach to analysing phraseology and collocation

in ESP texts.’ Anglais de Spécialité, 59, 5-23.

Goldberg, A.E. (1995). Constructions: A Construction Grammar Approach to Argument

Structure. Chicago and London: University of Chicago Press.

Goldberg, A.E. (2006). Constructions at Work: The Nature of Generalization in Language.

Oxford: Oxford University Press.

Goldberg, A.E. (2013). ‘Constructionist approaches.’ In T. Hoffman & G. Trousdale (Eds.),

The Oxford Handbook of Construction Grammar (pp. 15-31). Oxford: Oxford

University Press.

Granger, S., & Paquot, M. (2008). Disentangling the phraseological web. In S. Granger & F.

Meunier (Eds.), Phraseology: An Interdisciplinary Perspective (pp. 27-50).

Amsterdam: John Benjamins.

32

Gries, S.T. (2008). Phraseology and linguistic theory: A brief survey. In S. Granger & F.

Meunier (Eds.), Phraseology: An Interdisciplinary Perspective (pp. 3-25).

Amsterdam: John Benjamins.

Gries, S.T., & Stefanowitsch, A. (2004). Extending collostructional analysis: A corpus-based

perspectives on ‘alternations’. International Journal of Corpus Linguistics, 9(1), 97-

129.

Gries, S.T., & Wulff, S. (2005). Do foreign language learners also have constructions?

Evidence from priming, sorting and corpora. Annual Review of Cognitive Linguistics,

3, 182-200.

Groom, N., & Grieve, J. (2019). The evolution of a legal genre: Rhetorical moves in British

patent specifications, 1711 to 1860. In T. Fanego & P. Rodríguez-Puente (Eds.),

Corpus-based Research on Variation in English Legal Discourse (pp. 201–234).

Amsterdam: John Benjamins.

Hilpert, M. (2013). Constructional Change in English. Cambridge: Cambridge University

Press.

Hilpert, M. (2014). Construction Grammar and its Application to English. Edinburgh:

Edinburgh University Press.

Hilpert, M. (2018). Three open questions in diachronic construction grammar. In E. Coussé,

P. Andersson, & J. Olofsson (Eds.), Grammaticalization meets construction grammar.

[Constructional Approaches to Language 21] (pp. 21-39). Amsterdam: John

Benjamins.

Hiltunen, T. (2010). Grammar and Disciplinary Culture: A Corpus-Based Study.

Unpublished PhD Thesis, University of Helsinki, Helsinki, Finland.

Hoey, M. (2005) Lexical Priming: A New Theory of Words and Language. Abingdon:

Routledge.

Hoey, M., & O'Donnell, M.B. (2008). Lexicography, grammar, and textual position.

International Journal of Lexicography, 21(3), 293-309.

Hoey, M., & O'Donnell, M.B. (2015). Examining associations between lexis and textual

position in hard news stories, or according to a study by … In N. Groom, M. Charles

& S. John (Eds.), Corpora, Grammar and Discourse: In Honour of Susan Hunston

(pp. 117-144). Amsterdam: John Benjamins.

Hoffmann, T. (2017). Construction grammars. In B. Dancygier (Ed.), The Cambridge

Handbook of Cognitive Linguistics (pp. 310-329). Cambridge: Cambridge University

Press.

33

Hoffman, T., & Bergs, A. (2018). ‘A construction grammar approach to genre.’ CogniTextes

18. http://journals.openedition.org/cognitextes/1032 DOI: 10.4000/cognitextes.1032

Hoffman, T., & Trousdale, G. (Eds.), (2013). The Oxford Handbook of Construction

Grammar. Oxford: Oxford University Press.

Hunston, S. (2002). Corpora in Applied Linguistics. Cambridge: Cambridge University Press.

Hyland, K. (2008). As can be seen: Lexical bundles and disciplinary variation. English for

Specific Purposes, 27(1), 4-21.

Jaworska, S. (2016). Using a corpus-assisted discourse studies (CADS) approach to

investigate constructions of identities in media reporting surrounding mega sports

events: The case of the London Olympics 2012. In I.R. Lamond & L. Platt (Eds.),

Critical Event Studies. Leisure Studies in a Global Era (pp. 149-174). Basingstoke:

Palgrave Macmillan.

Jing-Schmidt, Z. (2017). What are they good for? A constructionist account of

counterfactuals in ordinary Chinese. Journal of Pragmatics, 113(1), 30-52.

Jing-Schmidt, Z. & X. Peng. (2016). The emergence of disjunction: A history of

constructionalization in Chinese. Cognitive Linguistics, 27(1), 101-136.

Kay, P. (2002). An informal sketch of a formal architecture for construction grammar.

Grammars, 5, 1-19.

Kay, P. (2004). Pragmatic aspects of grammatical constructions. In L.R. Horn & G. Ward

(Eds.), The Handbook of Pragmatics (pp. 675-700). Oxford: Blackwell.

Kerz, E., & Wiechmann, D. (2016). Second language construction learning: Investigating

domain-specific adaptation in advanced L2 production. Language and Cognition,

8(4), 533-565.

Koester, A. (2012). Corpora and workplace discourse. In K. Hyland, M.H. Chau & M.

Handford (Eds.), Corpus Applications in Applied Linguistics: Current Approaches

and Future Directions (pp. 47-64). London: Continuum.

Kopaczyk, J., & Tyrkkö, J. (Eds.) (2018). Applications of Pattern-Driven Methods in Corpus

Linguistics. Amsterdam: John Benjamins.

Lakoff, G. (1987). Women, Fire, and Dangerous Things: What Categories Reveal about the

Mind. Chicago, IL: University of Chicago Press.

Lambrecht, K. (1994). Information Structure and Sentence Form. A Theory of Topic, Focus,

and the Mental Representation of Discourse Referents. Cambridge: Cambridge

University Press.

34

Linell, P. (2009). Grammatical constructions in dialogue. In A. Bergs & G. Diewald (Eds.),

Contexts and constructions (pp. 97–110). Amsterdam: John Benjamins.

Michaelis, L. (2013). Sign-based construction grammar. In T. Hoffmann & G. Trousdale

(Eds.), The Oxford Handbook of Construction Grammar (pp. 133-152). Oxford:

Oxford University Press.

Michaelis, L., & Lambrecht, K. (1996). Toward a construction-based model of language

function: The case of nominal extraposition. Language, 72(2), 215-247.

Morton, T., Walsh, S., & O’Keeffe, A. (2011). Analyzing university spoken interaction: a

corpus linguistics/conversation analysis approach. International Journal of Corpus

Linguistics, 16(3), 326-345.

Nikiforidou, K., & Fischer, K. (2016). On the interaction of constructions with register and

genre. Constructions and Frames, 7(2), 137-147.

Oakey, D. (2008). The form and function of fixed collocational patterns in research articles in

different academic disciplines. Unpublished PhD Thesis, University of Leeds, Leeds,

UK.

Oakey, D. (2009). Fixed collocational patterns in isolexical and isotextual versions of a

corpus. In P. Baker (Ed.), Contemporary Corpus Linguistics (pp. 140-58). London:

Bloomsbury.

Orpin, D. (2005). Corpus Linguistics and Critical Discourse Analysis: Examining the

ideology of sleaze. International Journal of Corpus Linguistics, 10(1), 37-61.

Östman, J.-O. (2005). Construction discourse: A prolegomenon. In J.-O. Östman & M. Fried

(Eds.), Construction Grammars: Cognitive Grounding and Theoretical Extensions

(pp. 121–44). Amsterdam: John Benjamins.

Östman, J.-O., & Trousdale, G. (2013). Dialects, discourse, and Construction Grammar. In T.

Hoffmann & G. Trousdale (Eds.), The Oxford Handbook of Construction Grammar

(pp. 476-490). Oxford: Oxford University Press.

Partington, A., Duguid, A., & Taylor, C. (2013). Patterns and Meanings in Discourse:

Theory and Practice in Corpus-Assisted Discourse Studies (CADS). Amsterdam: John

Benjamins.

Patten, A. (2012). The English It-cleft: A Constructional Account and a Diachronic

Investigation. [Topics in English Linguistics 79]. Berlin: De Gruyter Mouton.

Perek, F. (2015). Argument Structure in Usage-Based Construction Grammar: Experimental

and Corpus-Based Perspectives. Amsterdam: John Benjamins.

35

Perek, F. (2016). Using distributional semantics to study syntactic productivity in diachrony:

A case study. Linguistics, 54(1), 149–188.

Potts, A., Bednarek, M., & Caple, H. (2015). How can computer-based methods help

researchers to investigate news values in large datasets? A corpus linguistic study of

the construction of newsworthiness in the reporting on Hurricane Katrina. Discourse

& Communication, 9(2), 149-172.

Römer, U., O’Donnell, M.B., & Ellis, N.C. (2015). ‘Using COBUILD grammar patterns for a

large-scale analysis of verb-argument constructions: Exploring corpus data and

speaker knowledge.’ In N. Groom, M. Charles & S. John (eds.). Corpora, Grammar

and Discourse: In Honour of Susan Hunston (pp. 43-71). Amsterdam: John

Benjamins.

Rühlemann, C. (2014). Narrative in English Conversation: A Corpus Analysis of Storytelling.

Cambridge: Cambridge University Press.

Rühlemann, C. (2018). Corpus Linguistics for Pragmatics: A Guide for Research. Abingdon:

Routledge.

Rühlemann, C., & Clancy, B. (2018). Corpus linguistics and pragmatics. In C. Ilie & N.R.

Norrick (Eds.), Pragmatics and its Interfaces (pp. 241-266). Amsterdam: John

Benjamins.

Sinclair, J.M. (1991). Corpus, Concordance, Collocation. Oxford: Oxford University Press.

Sinclair, J.M. (1996). The search for units of meaning. Textus, 9(1): 75–106.

Sinclair, J.M, (2004) Trust the Text: Language, Corpus and Discourse. R. Carter (Ed.).

London: Routledge.

Stefanowitsch, A., & Gries, S.T. (2003). Collostructions: Investigating the interaction

between words and constructions. International Journal of Corpus Linguistics, 8(2),

209-43.

Stubbs, M. (1997). Whorf’s children: Critical comments on critical discourse analysis (CDA).

In A. Ryan & A. Wray (Eds.), Evolving models of language (pp. 100-116). Clevedon:

Multilingual Matters.

Tadros, A. (1994). Predictive categories in expository text. In M. Coulthard (Ed.), Advances

in written text analysis (pp. 69-82). London: Routledge.

Taylor, C. & Marchi, A. (Eds.) (2018). Corpus Approaches to Discourse: A Critical Review.

Abingdon: Routledge.

Teubert, W. (2005). My version of corpus linguistics. International Journal of Corpus

Linguistics, 10(1), 1-13.

36

Teubert, W. (2010). Meaning, Discourse and Society. Cambridge: Cambridge University

Press.

Toolan, M. (1997). What is critical discourse analysis and why are people saying such

terrible things about it? Language and Literature, 6(2), 83-103.

Toolan, M. (2009). Narrative Progression in the Short Story: A Corpus Stylistic Approach.

Amsterdam: John Benjamins.

Traugott, E.C., & Trousdale, G. (2013). Constructionalization and Constructional Changes.

Oxford: Oxford University Press.

Upton, T.A., & Cohen, M.A. (2009). An approach to corpus-based discourse analysis: The

move analysis as example. Discourse Studies, 11(5), 585–605.

Wulff, S. (2008). Rethinking Idiomaticity: A Usage-based Approach. London: Continuum.

Wulff, S., Stefanowitsch, A., & Gries, S.T. (2007). Brutal Brits and persuasive Americans:

variety-specific meaning construction in the into-causative. In G. Radden, K. Köpcke,

T. Berg & P. Siemund (Eds.), Aspects of Meaning Construction (pp. 265-81).

Amsterdam: John Benjamins.

Yoon, J. & Gries, S.T. (2016). Corpus-based Approaches to Construction Grammar.

[Constructional Approaches to Language 19]. Amsterdam: John Benjamins.

Appendix 1: BNCweb ‘Distribution’ analysis results for the query way in which

(‘derived text type’ data only).

Derived text type:

Category No. of

words

No. of

hits

Dispersion

(over files)

Frequency

per

million words

Academic prose 15,778,028 1,292 321/497 81.89

Other spoken material 6,175,896 277 127/755 44.85

Non-academic prose and

biography 24,178,674 952 324/744 39.37

Other published written 17,924,109 545 179/710 30.41

37

material

Unpublished written material 4,466,673 131 63/251 29.33

Newspapers 9,412,174 123 78/486 13.07

Fiction and verse 16,143,913 103 68/452 6.38

Spoken conversation 4,233,962 15 10/153 3.54

total 98,313,429 3,438 1,170/4,048 34.97